Graph Based Recognition of Grid Pattern in Street Networks

摘要: Pattern recognition is an important step in map generalization. a street network significant for A grid characterized by set of mostly parallel lines, which are crossed second lines roughly at right angles. Inspired object image processing, this paper presents approach to based on graph theory. Firstly, bridges and isolated points the idenepsied repeatedly deleted. Secondly, similar orientations created, vertices represent segments edges orientation relationships between streets. Thirdly, candidates extracted through operators such as finding connected components, maximal complete sub-graphs, joins intersections. Finally, evaluated deleting reorganizing them into stroke models, changing these models intersection graphs strokes intersecting each other. The average clustering coefficient then calculated. Experimental results show that proposed valid detecting pattern lower degradation situations.